Shocovox is a Sparse Voxel Octree implementation in WGPU shaders ( hence: shady ). The leaf nodes contain 8 Voxel bricks instead of a single Voxel. This makes it possible to have a unique compression system, where Voxels of different resolutions can be mixed together. An implementation for raytracing is available with GPU support! The library uses Left handed Y up coordinate system.
- Implementing Caching to request data on demand to handle large data: https://github.com/davids91/shocovox/milestone/3
- Displaying large data as a panorama: https://github.com/davids91/shocovox/milestone/1
- Finalising "Octree of Voxel Bricks" concept: https://github.com/davids91/shocovox/milestone/2
These are the issues I will work on until 2025 Q2. I am eliminating them in a sequential manner.
- #56 - Introduce Palettes - Trimming the fat in Voxel storage, broadening the possibilities with user data and eliminating some data conversion overhead with bricks.
- #65 - Flatten brick storage: trimming some additional overhead, and eliminating some possible techDebt (
DIMgeneric argument with Octree) - #3 - to make it possible to have a limitless octree: so it's not bound by the RAM size anymore
- #17 Beam Optimization - Pre-render a small resolution image to optimally initialize ray distances, and help with deciding which bricks to load pre-emptively. GOTTA GO FAST
- #28, #6 - Level of Detail implementation to render large scenes more efficiently
I have marked some issues with the help needed flag, which I think would be a good addition to the library, but I can not focus on it as I am but a single person with limited time and resources. Feel free to try to tackle any of the marked issues (Or basically anything you'd like), I will provide any needed help and support if I can.
@nerdachse For the Albedo type and amazing support!
@DouglasDwyer My nemesis; Check out his project it's amazing! ( I hate him )
@Neo-Zhixing For his amazing project and awesome idea about how to utilize hardware RT for Voxel rendering!
